Why is my job taking twice as long?
We are running a series of overnight downloads from the iSeries using SQL server with ODBC connections. The job runs in subsystem QUSRWRK. The job, which once took two hours, leapt to a current five hours with no apparent changes to jobs, configuration or system parameters. The job ends normally, therefore there are no joblogs. On the odd occasion, jobs have to be manually terminated due to overrun. Do you have any idea what might be going on? I'm working with an iSeries Model 820 OS V5R2.
That may be the result of the SQL statement in the iSeries not optimizing (it can happen from changes in the database such as number of rows in tables, dropping indexes etc.).
I would recommend investigate into the QZDASOINIT jobs as follow:
- Force debug messages by setting parameter MESSAGES_DEBUG to *YES in the qaqqini file
- Force the joblog to print by temporarily changing the job description.
The joblog will show you the SQL used, how much time every statement taken, and the access plan (How did SQL planned to get the records).
This was first published in December 2005